Hardware isolation
Every sandbox boots its own guest kernel.
Security model
Each sandbox is its own Cloud Hypervisor VM on stock KVM. The isolation boundary is the hypervisor, not a container.
# on the control plane
curl -fsSL https://raw.githubusercontent.com/kortix-ai/platinum-sandbox/main/install.sh \
| bash -s -- --domain api.example.com --email you@example.com
# on the sandbox host: the line the installer prints, plus this host's IP
curl -fsSL -H 'Authorization: Bearer htok_...' https://api.example.com/install.sh \
| sudo bash -s -- htok_... https://api.example.com default default-1 10.0.0.6Replace api.example.com with your own domain and point its DNS at the control plane first. The quick start walks through both machines and your first API key.
Both SDKs read PT_API_URL and PT_TOKEN.
import { Platinum } from '@platinum-dev/sdk';
const sbx = await new Platinum().sandboxes.create({}, { waitForRunning: true });
console.log((await sbx.exec(['uname', '-r'])).stdout); // the guest's own kernel
await sbx.delete();from platinum import Platinum
sbx = Platinum().sandboxes.create(wait_for_running=True)
print(sbx.exec(["uname", "-r"]).stdout) # the guest's own kernel
sbx.delete()